Langue
< Home
Insights
Cutting Edge Cloud Technology teamed with Moments Lab’s Signature Multimodal Cognitive AI
By
Moments Lab Content Team
April 1, 2020
How Moments Lab (ex Newsbridge) is leveraging AWS cloud computing to build its proprietary and powerful AI indexing technology for media asset management.
Co-founded in 2016 by brothers Frédéric and Philippe Petitpont in Boulogne-Billancourt, France, Moments Lab (ex Newsbridge) is a cloud-based, AI-powered MAM platform for production teams. Leveraging its smart media asset management via facial, object and language analysis, Moments Lab offers journalists, production companies and rights holders immediate and auto-organized access to their most relevant content.
By bringing together artificial intelligence, cognitive technologies and extensive broadcast expertise, Moments Lab is revolutionizing the video production and editing processes for advanced and real-time story production via an automated media logging solution, processing several thousand hours of content each month for the largest European publishers, television channels (M6 Group, TF1 Group) and sports rights holders (adidas, Evian Championship or French Rugby Federation).
Each day, production teams consolidate and upload mass amounts of content to digital databases which then needs to be analyzed by human beings. From categorizing assets to identifying sources, scenes, themes, locations, personalities, etc. the job is becoming an increasingly demanding task with little room for error.
Additionally, publishing time continues to be a production challenge, as time is money in the industry. Most news outlets prioritize what is hot off the press, with journalists who take their time fact-checking and understanding material often left in the dust. Instead, newsrooms take quickly populated and not-so-relevant stories, resulting in lower quality deliverables.
Having a background in broadcast engineering and computer science with a shared mission to enhance advanced story production via smart digital asset management, the Petitpont brothers had a revolutionary idea, and soon-after the Moments Lab platform was conceived.
With a core objective of expediting the post-production process resulting in lightning-speed with near-live publish times (with cut costs as an added bonus), Moments Lab’s AI-Based video indexing MAM platform leverages both Moments Lab AI technology and the scaling capacity of AWS computing infrastructures. Such technology includes but is not limited to: facial recognition, object detection, speech-to-text and context analysis. Unique to competitors, Moments Lab is an all-in-one digital asset management platform that smartly indexes assets in real-time (compared to a simplistic transcription or digital asset storage services).
Frédéric Petitpont, Moments Lab CTO and Co-founder: "Moments Lab solutions are powered by our multimodal cognitive AI teamed with Amazon Comprehend and Amazon EC2's latest Nvidia's GPU instances to deliver the best video asset management experience to media-led organizations worldwide."
At Moments Lab (ex Newsbridge), user experience and product quality are paramount. With a designated UX department and expert collaboration with Amazon Web Services, harnessing the elements of platform design and high-level functionality is a strong suit within the organization. From the beginning, these two components were prioritized as the core value of Moments Lab’s AI-powered platform lies in the simplification of the advanced investigation process.
In order to make this objective a reality, the Moments Lab platform has been designed to include an array of proprietary, complex and powerful, existing datasets. These data sets are part of a big data framework which the Moments Lab team has adapted for video, with microservices at an infinite scale.
With the added-value of Amazon expertise, Moments Lab uses its signature multimodal AI technology in collaboration with the following AWS capabilities to carry-out the smart-indexing of hundreds of thousands of digital media assets in real-time, using a variety of microservice algorithms.
o Amazon EC2 G4 Instances run neural networks and machine learning applications to train and detect patterns from custom datasets imported by customers. Moments Lab is able to harness the power of Nvidia GPU and the very latest instances g4dn to cut costs and compute advanced multimodal inferences (crossing transcription, patterns, facial and object recognition to achieve an improved confidence level).
Moments Lab decided to initiate collaboration with AWS for several reasons. First, Amazon Web Services are highly available, flexible and scalable as applied to our infrastructure.
Next, state-of-the-art technologies are also able to operate with and power a multitude of components.
Additionally, autonomy was another factor. Amazon’s “everything available + self service” policy was the kind of service we wanted, as it’s aligned with our foundational objective of empowering journalists, editors and rights holders with their media.
Finally Transparency, certifications available on demand as compliance and privacy are not a selling point but a prime requirement for Moments Lab and one of the ingredients of trust with our clients.
As previously mentioned, the Moments Lab platform is a (very) smart media asset management system with advanced yet secure organizational indexing capabilities.
To give you a clear idea, once a user logs into the platform he or she lands on the homepage dashboard, in which a centralized cognitive search engine can be found at the top of the screen, with mission streams organized by AI below.
In order to populate these streams, users are able to create a live or file-based mission and then select the outbound destination. Once the mission is created, the platform immediately and automatically analyzes, and indexes the given media asset.
To implement the components of Moments Lab Platform, the initiative was spearheaded with several implementation phases and a combination of AWS Cloud Services along with proprietary Moments Lab Multimodal AI with several open source integrations.
AWS EC2 G4 Instances is the infrastructure used to perform user’s custom pattern recognition and multimodal inferences over live streams and video files.
This infrastructure works by taking a heavy video file or live stream which needs indexed, and processing it through a network of multimodal actors (we could see these as autonomous microservices) which can range from speech-to-text to facial recognition for example- creating an output of video material that is compressed, smartly indexed and stored.
The system orchestrating each AI component is called Akka, implemented using the Scala language, which helps each microservice work with or “communicate” with the others. This technology is especially helpful when dealing with a very high number of threads in parallel, as it creates an actor system disptached over a cluster of instances. Instances are set in an EC2 autoscaling group to ensure every task is processed faster than real time. Cognitive analysis performed like this is way more efficient than a traditional sequential linear model (without multimodal analysis capabilities) that can often be hardwared-capped when deployed on-prem - it is actually the only way to have multimodal cognitive analysis done live for a virtually unlimited amount of streams. This saves producers time and money along with creating extreme asset value and freshness.
Through the unique cloud-based infrastructure on AWS, the Moments Lab Platform enables production teams to quickly, smartly and accurately index and manage their digital media assets. This has resulted in accelerated digital media asset management for numerous production teams in an array of industries. Additionally, the customizability and agility of the platform provides to the storytelling community the ability to quickly and efficiently search for, analyze and categorize assets via a real-time solution, based on their self-populated, complete pool of assets.
The result is an off-the shelf, comprehensive solution that adds clear value to story-tellers and producers world-wide.
Amazon Web Services remains a trusted provider, as Moments Lab will harness AWS worldwide locations and availability to deliver services close to its international clients’ offices.
.png)
